ABOVE: MICROBIAL CONNECTIONS: Nanotubes can be seen extending from E. coli in this scanning electron micrograph of a coculture of E. coli and Acinetobacter baylyi.

In 2011, the microbiology community learned of a brand-new feature of bacteria: nanotubes. Scientists later showed that these membranous, hollow connections between bacteria allow the transfer of materials such as amino acids and toxins that inhibit growth. These tubes were unlike anything the researchers had seen before: in contrast to the conjugative pili that transfer genetic material during bacterial “sex,” the nanotubes were made of lipids, not proteins. They were also more promiscuous than pili, often linking microbes of different species, and even connecting bacteria with mammalian cells. It was starting to look like bacterial nanotubes were long-overlooked features of microbiology.

Jiří Pospíšil, a graduate student at the Czech Academy of Sciences, was enamored with these novel...

Not one to give up, Pospíšil struck up a collaboration with Imrich Barák of the Slovak Academy of Sciences. Barák had access to more sophisticated imaging technology, and Pospíšil says he hoped that he could use it to capture the elusive structures. Together, the researchers prepared Bacillus subtilis for microscopy using glass slides and coverslips, standard tools of the trade. One day, they saw that the cells were moving on the slide. To immobilize them, “we just decided to push the glass coverslip down,” says Barák. When they did, lo and behold, nanotubes began bursting forth from nearly every cell. Looking closer, Pospíšil and Barák found that the cells were dying. 

This observation, published last October, ran contrary to all the previously published work on nanotubes, which posited these structures as conduits between living, healthy cells. If nanotubes were indeed features of dying cells, then the suggestion that the structures were central to material transfers that seemed to play a role in microbial growth and survival was unlikely to hold true, says Barák. “We realized that probably this is the end of this nice story of different functional nanotubes.”

What are bacterial nanotubes?

As with many scientific advances, bacterial nanotubes were discovered by accident. In 2008, Sigal Ben-Yehuda, a microbiologist at the Hebrew University of Jerusalem in Israel, and her then–graduate student Gyanendra Dubey were studying sporulation in B. subtilis. One of their experiments involved mixing two types of B. subtilis, one that expressed green fluorescent protein (GFP) and one that did not, and letting them grow together. When they observed the cells under a light microscope, they noticed weak fluorescence coming from some of the cells that originally did not have GFP but were lying close to cells that did. “I thought that [it was] a bleaching artifact,” says Dubey, who until recently worked as a researcher at the Institut Pasteur. 

Ben-Yehuda and Dubey performed a battery of tests to figure out what was going on. They did experiments using other fluorescent small molecules, such as calcein, and found that these also seemed to jump between cells, as did genetic material such as plasmids or RNA. Bacteria can exchange material using contact-independent systems, such as simple diffusion and vesicles, or contact-dependent systems, such as conjugative pili and secretion systems. But the scientists observed material-sharing even in strains of B. subtilis that were incapable of these well-known modes of transfer. “That made us think that there was some cytoplasmic exchange happening that we were [previously] not aware of,” says Ben-Yehuda.

We realized that probably this is the end of this nice story of different functional nanotubes.

—Imrich Barák, Slovak Academy of Sciences

When she and Dubey examined B. subtilis with high-resolution scanning electron microscopy (HR-SEM), they saw tubes extending from the bacteria to adjacent cells. “We could see them again and again and again,” says Ben-Yehuda. Surprised that these structures had not been reported before, they dug through older papers for electron microscope (EM) images of bacteria. To their delight, they “could see them all over the place all the time,” says Ben-Yehuda. She surmised that bacterial nanotubes had been overlooked simply because no one really knew what they were. In 2011, the team published its first paper reporting the existence of bacterial nanotubes that acted as channels connecting bacterial cells.

Until the discovery of bacterial nanotubes, the only known contact-dependent systems that could transfer material between bacterial cells were made of proteins, and researchers had shown these conduits to be very picky about the cells they connect to; conjugative pili, for instance, will only form between bacteria that carry specific plasmids and others that carry certain receptors. Ben-Yehuda’s group found that, unlike these protein-based systems, nanotubes are made almost entirely of lipids and do not seem to care about who they partner up with. Often looking like a string of connected vesicles, nanotubes vary in width and facilitate cytoplasmic connections between cells. The cytoplasmic link distinguishes nanotubes from outer membrane vesicles that package and secrete material from the space between the two membranes of Gram-negative bacteria.

Curious about how widespread bacterial nanotubes are, Ben-Yehuda’s team tested various other species of bacteria. The lab next door, led by Ilan Rosenshine, worked on enteropathogenic E. coli (EPEC), which attaches to its host cells to extract nutrition through hitherto unknown molecular machinery. Ben-Yehuda thought that it might be nanotubes. She approached Rosenshine, and asked him to share some EPEC samples with her. In 2019, they published two papers together, one reporting that EPEC did indeed produce nanotubes when attaching to host cells, and the other showing that nanotubes can connect Gram-positive and Gram-negative bacteria, and that the microbes require a set of five proteins, called the CORE complex, to do so. Neither EPEC nor B. subtilis cells that lacked this complex could produce nanotubes. “We couldn’t believe that all these different systems, from EPEC and from Bacillus, were the same,” says Ben-Yehuda. 

Meanwhile, several other labs were also hard at work documenting the presence of nanotubes in their own bacterial systems. In 2015, Christian Kost of the University of Osnabrück in Germany and his colleagues engineered two species of bacteria, E. coli and Acinetobacter baylyi, to each lack a specific amino acid. Paired populations of the microbes, with each of the pair lacking a different amino acid, would only survive if they grew together and exchanged amino acids. When the researchers tagged amino acids with fluorescent markers and observed them under a microscope, they saw that the amino acids were exchanged between the two cell types. They tracked the movement of these fluorescent markers over time and found that the transfer was happening through lipid-based tubes. Using EM, the scientists then confirmed that these were bacterial nanotubes.

That same year, a paper from Marie-Thérèse Giudici-Orticoni’s group at the French National Centre for Scientific Research and Aix-Marseille University reported seeing nanotube-like structures connecting Clostridium acetobutylicum, a Gram-positive bacterium, and Desulfovibrio vulgaris, which is Gram-negative. The contact between cells, which they observed using scanning electron microscopy (SEM) and other techniques, appeared to allow the cells to exchange molecules in order to survive harsh growth conditions, such as a shortage of nutrients.

More recently, in mid-2020, Jinju Chen’s group at Newcastle University documented the presence of nanotubes in Pseudomonas aeruginosa, a common, biofilm-forming pathogen found in hospitals. Faced with nanopillars—tiny spikes that are often used to make medical devices with surfaces that can deter the formation of harmful biofilms—the bacteria aligned themselves between the pillars and used nanotubes to connect different cells. By using strains that were incapable of forming other contact-enabling structures such as pili, the group confirmed that the connecting structures they saw were indeed nanotubes. They concluded that the nanotubes might be making it easier for the bacteria to communicate, possibly promoting the cells’ ability to form biofilms in hostile environments.

With so many labs having observed bacterial nanotubes under different conditions and in actively growing cells, Pospíšil was not anticipating that he would have any trouble doing the same.

Elusive structures

When Pospíšil first set out to visualize bacterial nanotubes, he immediately noticed that the structures occurred at a much lower frequency than had been reported by Ben-Yehuda’s lab. “The greatest challenge to study nanotubes was actually to be able to detect nanotubes,” says Libor Krásný, Pospíšil’s advisor at the Czech Academy of Sciences in Prague. In fact, the researchers only found one nanotube for every 500 or so cells they scanned. 

“We [saw] something which looks like nanotubes, but in very, very low frequencies,” says Oldřich Benada, head of the EM group in the Institute of Microbiology of the Czech Academy. “And we had to scan a lot of frames to find some of those structures.” Benada says he grew skeptical that nanotubes existed at all.

A couple of years into their exploration of nanotubes, Krásný met Ben-Yehuda at a conference and discussed his group’s difficulties observing the structures. Ben-Yehuda later provided them with the same bacterial strains and protocols her lab had used, but the Czech group remained unable to see the nanotubes. The researchers also tried and failed to reproduce some of the material transfer experiments that Ben-Yehuda’s lab had performed. 

That was when Pospíšil and his Slovak Academy of Sciences collaborator Barák started applying pressure to their coverslips and inadvertently discovered that this made the bacteria produce lots of nanotubes. They immediately started testing varying amounts of pressure to see what could most reliably make the cells produce nanotubes. They found that applying about 80 kilopascals, which they achieved by placing a 2.5-kilogram weight on the coverslip for 10 seconds, almost immediately resulted in several nanotubes extending from each cell. 

When Pospíšil used a marker called SYTOX Green, which only stains dead or dying cells, he saw that nanotubes were exclusively produced by green cells. This told him that the pressure was killing the cells, which produced nanotubes as a result. Even though pressing down on the coverslip is a common practice, Barák says he thinks that pressure-induced nanotubes have never been reported before because people tend to ignore dead cells. He adds that in decades of using dyes that stain cell membranes and pressing on coverslips, “I didn’t ever see nanotubes . . . because I didn’t concentrate on them.” Now, he knows that if he carefully fixes the cells and uses a good microscope, he will likely see nanotubes all over the place.

What Are Bacterial Nanotubes? 

Unlike cellular appendages such as the pili used for mating, injectisomes that transfer virulence proteins, and flagella that power swimming in many microbes, bacterial nanotubes are made solely of lipids and can connect the cytoplasm of different microbial species.

Bacterial Nanotubes

Conjugative Pili

Type 3 Secretion Systems, e.g., Injectisomes and Flagella

Composition; structure

Lipids; segmented

Proteins; helical

Multiprotein complex; tubular


1–40 μm

1–2 μm

0.8–2 μm


30–130 nm; commonly 40–70 nm

6–11 nm; lumen diameter ~3 nm

8–10 nm; lumen diameter ~2.5 nm

Materials transferred

Antibiotic resistance factors, metabolites, toxins


Injectisomes for the transfer of virulence proteins; flagella for motility

Proteins involved in formation

CORE complex (same proteins as the flagellar base) and hydrolases that help make a hole in the cell wall

“Transfer” (Tra) class of proteins such as pilin, TraL, and TraF

The injectisome complex has various proteins such as secretin, stalk protein, and needle filament; the flagellar apparatus has its own set of dedicated proteins that form the base, stalk, and tip.

See full infographic: WEB | PDF

When the group published its work in December 2020, Barák says, the response from the scientific community was largely positive. The paper was shared widely on Twitter, drawing comments from a few scientists who questioned whether the nanotubes were indeed real structures. 

Alex Merz, a biochemist at the University of Washington, has been a vocal critic of some of the papers from Ben-Yehuda’s lab, especially the recent one detailing the involvement of the CORE complex and showing that nanotubes can connect Gram-positive and Gram-negative bacteria. These two sets of bacteria have very different cell envelopes—Gram-positive bacteria have a single lipid membrane covered by a tough peptidoglycan cell wall, whereas Gram-negative bacteria have an additional lipid membrane outside of a thinner cell wall. When a nanotube extends from a Gram-positive bacterium toward a Gram-negative bacterium, does it interact only with the outer membrane of the Gram-negative recipient, or does it actively engage with the inner membrane of the recipient cell? Merz says he believes that answering these questions is vital to figuring out whether bacterial nanotubes are indeed functional structures or artifacts. He calls for the use of higher-resolution microscopy techniques, which have not been used in the recent papers, that would allow one to closely observe how the nanotubes interface with the two types of bacterial cell envelopes. “I need to see more to be persuaded,” he says. 

Erin Goley, a biological chemist at Johns Hopkins University School of Medicine, echoes Merz’s concerns. “I am skeptical of them being real, functional structures,” she says. Aside from how the nanotube membrane is organized, she is puzzled by the seemingly generalist nature of nanotubes. Moreover, as nearly every contact-dependent system seen so far has its own tightly regulated machinery, Goley finds it strange that nanotubes seem to lack a master regulator. Like Merz, she is interested in seeing high-quality cross-sectional images of the nanotubes, particularly at the points where they contact the bacterial cells. The kinds of techniques both Goley and Merz recommend will let scientists eliminate traditional sample preparation artifacts and preserve cells in their native states, which would help “show you actually what’s going on with the membranes . . . who’s contacting what and which membranes are contiguous with which,” says Goley. 

Kost says he thinks that the pressure-induced nanotubes seen by the Czech group might reflect the self-organization of lipids once they burst out of dead cells, similar to an older study showing that protocells, self-organizing lipid vesicles capable of interacting with their environments, can form long lipid tubes under certain conditions. “I think, yes, under some conditions, cells show this phenomenon. But this certainly does not rule out nanotubes” playing a role in living cells, he adds.

Ben-Yehuda’s group stands by its work showing the common- place formation of nanotubes among living microbes. “We have never had to place the cells under any sort of pressure,” says Amit Baidya, a postdoc in the lab. Ben-Yehuda also notes that bacterial colonies that they imaged were still capable of growing and dividing after having made nanotubes. 

Even in her lab, however, the researchers sometimes struggle to observe nanotubes. A major challenge is the lack of tags for nanotubes, forcing scientists to rely on membrane staining and resulting in very noisy images. “It’s like looking at the sun and trying to see something very small,” says Ben-Yehuda. Baidya agrees that the process of identifying nanotubes is finicky. “You have to be really, really careful while making the samples,” he says. “If you are not focused that day . . . if you are in a hurry, you will not see nanotubes.” 

Sources of Variation in Nanotube Studies

To interrogate bacterial nanotubes, researchers first grow the bacteria in culture, then use sophisticated microscopy techniques and different readouts such as fluorescent tags to check whether cells are exchanging materials. The behavior and morphology of bacterial cells can differ depending on the specifics of each step, giving rise to varied observations and, sometimes, conflicting results.

Step 1: Grow the bacteria

Bacteria can be grown either on solid media or in liquid broth. Depending on the species, growth conditions can influence bacterial behavior and appearance.

Bacteria grown in both types of media have been seen to produce nanotubes, but some labs have been unable to observe them in either one type or the other.

Step 2: Test for the transfer of materials between two bacteria

Observing whole cells or colonies can reveal the transfer of fluorescent proteins, plasmids, metabolites, and more.

Various methods have been used to observe nanotubes, but some labs have been unable to observe certain types of material transfer.

Step 3: Prepare the bacteria for imaging

For fluorescence microscopy, bacteria are either made to express fluorescent proteins or are treated with membrane-staining dyes. Glass slides and coverslips can be coated with different compounds to allow the bacteria to stick better. For electron microscopy, cells are dried and treated with heavy metals before imaging.

Differences in sample preparation can affect the reliability with which bacterial nanotubes are observed. One lab, for example, could only see nanotubes if they pressed down on the coverslip before imaging the cells, whereas other labs never needed to do so in order to observe the structures.  

Step 4: Image the bacterial nanotubes

Fluorescence and electron microscopy, or sometimes a combination of both, are most commonly used to observe nanotubes. The structures are hard to see using fluorescence microscopy (right), a lower-resolution form of imaging, because there are no specific markers for nanotubes. Using higher magnification electron microscopy (left), nanotubes can be easily resolved and distinguished from structures such as pili and flagella.

Nanotubes can only be identified with training and experience; scientists need to be able to distinguish them from the background noise of fluorescence microscopy and artifacts of electron microscopy.

See full infographic: WEB | PDF

A replication crisis?

The studies on bacterial nanotubes differ in many aspects: the bacterial species or strains, growth conditions, assays to detect material transfer, and microscopy techniques to image the structures themselves. And, as Pospíšil’s study highlighted, not all labs have success with all the conditions. 

The growth conditions of bacteria, including temperature, acidity, and aeration, are known to affect the physiology and appearance of cells, says Paula Montero Llopis, who directs the light microscopy facility at Harvard Medical School. For example, B. subtilis comes in two forms: swarmers, which can freely swim around, and chains, which are made of cells that are connected to one another and therefore cannot move, explains Montero Llopis, whose work has revealed that the latter are more likely to form when the bacteria are grown at a low temperature in media containing casein hydrolysate, a source of amino acids. “I found that the percentage of swarmers vs. chains in the population is highly dependent on how you grow the cells,” she tells The Scientist in an email. 

Putative Functions of Nanotubes

Numerous studies have identified various possible roles for bacterial nanotubes, which researchers have observed under different growth conditions.


Species found

Culture conditions


Transfer of materials (RNA, proteins, amino acids, toxins)

Bacillus megaterium, B. subtilis, Clostridium acetobutylicum*, Desulfovibrio vulgaris*, E. coli,
enteropathogenic E. coli, Staphylococcus aureus

Solid or liquid media, depending on the study

Cell, 144:590–600, 2011; Nat Comm, 11:4963, 2020; Nat Comm, 6:6283, 2015; Cell, 177:683–96.e18, 2019

Adhesion to mammalian cells

Enteropathogenic E. coli

Liquid media

Cell, 177:683–96.e18, 2019

Adhesion to nanopillars and the formation of biofilms

Pseudomonas aeruginosa

Liquid media

Soft Matter, 16:7613–23, 2020

Stress response in dying cells placed under pressure

B. subtilis

Liquid media

Nat Comm, 11:4963, 2020

*Indirect evidence

When it comes to nanotubes, Pospíšil and Kost both used liquid media, whereas Ben-Yehuda’s lab grew the bacteria on solid agar plates. Neither Pospíšil nor Kost was able to observe nanotubes from bacteria grown on solid media, while Ben-Yehuda’s lab failed to see them on cells in liquid media. Kost attributes different experimental outcomes to the bacterial species each lab handles: Ben-Yehuda’s lab works on B. subtilis, which is a soil bacterium and prefers solid surfaces, whereas Kost’s lab works on E. coli, which likes liquid cultures. 

Even when labs are trying to replicate one another’s studies, comparing results will be tricky, Kost notes. “We try to replicate all the materials that we use as good as possible, but you never know,” he says, having faced similar problems when shifting his own lab from one university to another. Not only is there variation among manufacturers of various reagents, Kost explains, but batch differences among chemicals can foil replication attempts even when using otherwise identical products for growing or imaging bacteria. 

These differences can also spill over into how the scientists prepare the bacteria for imaging and the kind of imaging technique used. When preparing bacteria for EM, two of the most common steps are drying the cells out and then coating them with metals. “These techniques . . . can be really robust, but they do carry a significant hazard of artifacts,” says Merz. He also cautions against chemical fixation, a technique that uses chemicals such as paraformaldehyde to “freeze” cells in time, as it is more likely to give rise to artifacts. Among the labs that have observed nanotubes using EM, the imaging modalities and sample preparation techniques used have varied widely. 

I am skeptical of them being real, functional structures.

—Erin Goley, Johns Hopkins University School of Medicine

The different and sometimes contrasting conditions required to observe nanotubes are not lost on nanotube skeptics. Ariane Briegel, a professor of ultrastructural biology at Leiden University, says she believes that the structures seen in EMs might in fact be drying artifacts, which is why they cannot be observed consistently. “I do believe that there is exchange [of materials] between the cells, but I am not sure how far I trust the nanotubes to be real,” she tells The Scientist in an email. Goley concurs, adding that the lack of replicability has only added to her skepticism. 

But to Kost, the fact that a variety of conditions have been used to produce nanotubes is evidence that they do indeed exist. “What I found very encouraging is that . . . there’s so many labs that report very, very similar phenomena.” In fact, the apparent ubiquity of nanotubes has spurred him to push forward. In 2019, his group found that the transport of amino acids through bacterial nanotubes was unidirectional—nanotubes that originated from bacterium A would shuttle materials from bacterium A to bacterium B, but not the other way around. This implies that the nanotubes provide an element of control and are not just tubes connecting the cytoplasms of two cells.

Kost says that ongoing studies in his lab indicate that the nanotubes might not just be mediators, but key instigators of bacterial cooperation. In unpublished work, the lab has observed that the mere presence of nanotubes makes cells choose cooperative strategies over competition, he says. “This completely changed the way that I look at ecological interactions. This seems to suggest that these nanotubes are the key for cooperation to evolve.”

Ben-Yehuda’s group, meanwhile, has continued to unearth details of nanotube formation. The researchers recently discovered, for example, that nanotube-producing bacteria first send to recipient cells hydrolase activators, which trigger cell wall remodeling to allow the nanotubes to penetrate and send other material through. Somehow, cells also appear to direct where the nanotubes go, Ben-Yehuda says. “They can grow everywhere in the space, no? [But] they always grow directly towards each other. . . . That makes me suggest that it’s not random.”

Pospíšil does not see this sort of directionality in the nanotubes extending from his dying bacteria, but he has observed a preference in the regions from which the nanotubes emerge. In vertically elongated B. subtilis cells, Pospíšil has observed that the nanotubes are produced mostly from the poles of the cells (the top and bottom), rather than from along the sides.

Tim Errington, the director of research at the Center for Open Science (COS), point outs that contrasting results are not always bad. “Different points of view [are] exactly what you want in science, because you shouldn’t [just] accept something you’re saying. . . . You have to keep revisiting it.” One of the ways Errington thinks that the differences regarding bacterial nanotubes can be resolved is through a method known as adversarial collaboration, where two labs with opposing hypotheses “precommit” to a set of experiments designed to test replicability, do the experiments, and then compare results.  

Ben-Yehuda remains unperturbed by Pospíšil’s findings and is excited to explore the myriad unanswered questions about bacterial nanotubes. “Over the years, we established that they are real and they are there, although some people think they are not. Science is a marathon, and we are running a marathon, not a sprint.”

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?